Machine learning for medical image reconstruction with phase correction

Abstract

For reconstruction in medical imaging using phase correction, a machine learning model is trained for reconstruction of an image. The reconstruction may be for a sequence without repetitions or may be for a sequence with repetitions. Where repetitions are used, rather than using just a loss for that repetition in training, the loss based on an aggregation of images reconstructed from multiple repetitions may used to train the machine learning model. In either approach, a phase correction is applied in machine training. A phase map is extracted from output of the model in training or extracted from the ground truth of the training data. The phase correction, based on the phase map, is applied to the ground truth and/or the output of the model in training. The resulting machine-learned model may better reconstruct an image as a result of having been trained using phase correction.

What is claimed is: 
     
         1 . A method of machine training for magnetic resonance (MR) reconstruction in medical imaging, the method comprising:
 acquiring MR training data including ground truth representations;   machine training a neural network for the MR reconstruction using the MR training data, wherein an output of the neural network and/or the ground truth representations are phase corrected; and   storing the neural network as machine trained.   
     
     
         2 . The method of  claim 1  wherein acquiring comprises acquiring the MR training data for an MR protocol using multiple repetitions and wherein machine training comprises training the neural network to output an image for each one of the multiple repetitions, a first loss used in the training being based on an aggregation of the images from the multiple repetitions. 
     
     
         3 . The method of  claim 2  wherein the aggregation is an average in a complex-value domain where the first loss comprises a combination of a complex-value loss and a magnitude-based loss. 
     
     
         4 . The method of  claim 1  wherein machine training comprises machine training where the phase correction is derived from the MR training data and applied to the output. 
     
     
         5 . The method of  claim 1  wherein machine training comprises machine training where the phase correction is derived from the output. 
     
     
         6 . The method of  claim 1  wherein machine training comprises machine training where the phase correction is derived from the MR training data and applied to the ground truth representations. 
     
     
         7 . The method of  claim 1  wherein machine training comprises applying the phase correction to the output and the ground truth representations. 
     
     
         8 . A method for reconstruction of a medical image in a medical imaging system, the method comprising:
 scanning, by the medical imaging system, a patient, the scanning resulting in scan data;   reconstructing, by an image processor applying a machine-learned model, a scan image, the machine-learned model having been trained with application of phase correction; and   displaying the medical image based on the scan image.   
     
     
         9 . The method of  claim 8  wherein the scan data comprises measurements over a series of scans of an imaging protocol, wherein reconstructing comprises reconstructing the scan image for each of the scans of the series, the machine-learned model having been trained for use for each scan of the imaging protocol based on a loss function from a combination of training images from different scans for the imaging protocol; and
 further comprising combining the scan images into the medical image. 
 
     
     
         10 . A system for reconstruction in medical imaging, the system comprising:
 a medical scanner configured to repetitively scan a region of a patient pursuant to a protocol, the scan providing scan data in repetitions of the protocol;   an image processor configured to reconstruct, for each of the repetitions, a representation of the region, the image processor configured to reconstruct by application of a machine-learned model having been trained for the reconstruction for each of the repetitions based on a loss function between an aggregate of outputs from the repetitions of the protocol and a ground truth, the loss function having been a complex-valued loss based on the output for each of the repetitions or component images of the ground truth being phase corrected, the image processor further configured to combine the representations from the repetitions; and   a display configured to display an image of the region from the combined representations.   
     
     
         11 . The system of  claim 10  wherein the machine-learned model was trained with the phase corrections for the outputs having been derived from the component images of the ground truth. 
     
     
         12 . The system of  claim 11  wherein the machine-learned model was trained with the phase corrections having been derived from the component images by low pass filtering and extraction from results of the low pass filtering. 
     
     
         13 . The system of  claim 10  wherein the ground truth comprised an aggregated ground truth.